JP2003517412A - Method for growing defect-free silicon crystals with arbitrarily large diameter - Google Patents

Abstract

(57) [Abstract] A method of growing a single crystal silicon ingot substantially free of coherent intrinsic point defects. Ingots are generally grown by the Czochralski method. While the ingot is grown, also without cooling any portion of the ingot to a temperature lower than the temperature T _A which aggregation occurs intrinsic point defects in the ingot. Thus, the production of defect-free ingots is substantially decoupled from process parameters such as take-off speed and system parameters such as axial temperature gradients in the ingot.

Description

Detailed Description of the Invention

TECHNICAL FIELD The present invention relates generally to the manufacture of semiconductor grade single crystal silicon used in the manufacture of electronic components. More specifically, the present invention provides an agglomerated intrinsic point defect in the total radius of the crystal and in the usable length of the ingot.
The present invention relates to a method for producing a single crystal silicon ingot substantially free from

BACKGROUND OF THE INVENTION Single crystal silicon, which is the starting material for most manufacturing processes for semiconductor electronic components, is generally
It is manufactured by the so-called Czochralski (Cz) method. In this way,
Polycrystalline silicon (polysilicon) is loaded into a crucible, melted, a seed crystal is brought into contact with molten silicon, and a single crystal is grown by slow pulling. After the neck formation is complete, the crystal diameter is increased until the desired or target diameter is reached by decreasing the pull rate and / or the melting temperature.
Next, while supplementing the decreasing melt volume, the pull rate and melt temperature are adjusted to grow a tubular body of crystals having a substantially constant diameter. Near the end of the growth process, the crystal diameter must be gradually reduced to form an end-cone before the crucible is depleted of molten silicon. End cones are generally formed by increasing the crystal pull rate and the heat supplied to the crucible. When the diameter is small enough, the crystals separate from the melt.

Recently, many defects in single crystal silicon have been found to form in the crystal growth chamber as the crystal cools after solidification. Such defects result, in part, from the presence of excess (ie, concentrations above the solubility limit) intrinsic point defects in the crystal lattice that are vacancies and self-interstitials. Silicon crystals grown from the melt generally grow with an excess of one type of intrinsic point defect, either a crystal lattice vacancy (V) or a silicon self-interstitial (I). The type and initial concentration of these point defects in silicon are determined during solidification, and when these concentrations reach the level of critical supersaturation in the system and the point defect mobility is sufficiently high, a reaction or aggregation event occurs. It has been reported that this can happen. Agglomerated intrinsic point defects in silicon have a significant impact on the yield capability of materials in the fabrication of complex, highly integrated circuits.

Vacancy type defects are observable crystals such as D defects, flow pattern defects (FPD), gate oxide integrity (GOI) defects, crystalline origin particle (COP) defects, and crystalline origin light point defects (LPD). It is believed to be the source of defects and certain bulk defects observed by infrared scattering methods such as scanning infrared microscopy and laser scanning tomography. Ring oxidation induced stacking fault (O
Defects that act as nuclei of ISF) also exist in the excess vacancy region. This particular defect is believed to be a hot nucleated oxygen agglomerate catalyzed by the presence of excess vacancies.

Defects associated with self-interstitial atoms have not been well studied. They are generally considered to be low density interstitial dislocation loops or networks. Although such defects do not contribute to gate oxide integrity defects, which are an important wafer performance criterion, it is widely recognized that they are the cause of other types of device defects commonly associated with leakage current problems. There is.

The density of such vacancies and self-interstitial agglomeration defects in Czochralski silicon is generally in the range of about 1 * 10 ³ / cm ³ to about 1 * 10 ⁷ / cm ³ .
Although these values are relatively low, agglomerated intrinsic point defects are becoming an increasingly serious problem for device manufacturers and are in fact considered to be the yield limiting factor in the device manufacturing process.

Currently, there are three main approaches to address the problem of agglomerated intrinsic point defects. The first method involves focusing on the crystal pulling method to reduce the number density of agglomerated intrinsic point defects in the ingot. This method can be further classified into a method having a crystal pulling condition for forming a vacancy dominant material and a method having a crystal pulling condition for forming a self-interstitial dominant material. For example, crystals adjust the v / G _O to grow, and (ii) during crystallization taking over step, from about 1100 ° C. to about 1050 ° C. is (i) predominant intrinsic point defect crystal lattice vacancies are It has been reported that the number density of agglomerated defects can be reduced by affecting the nucleation rate of agglomerated defects by changing (generally decreasing) the cooling rate of the silicon ingot. Although this method reduces the number density of agglomerate defects, it cannot prevent the formation of agglomerate defects. The presence of these defects is becoming a more serious problem as device manufacturers are becoming more demanding.

It has also been proposed to reduce the take-off rate to a value of less than about 0.4 mm / min during the growth of the crystal body. However, such a slow pulling rate also reduces the throughput of the nuclear puller, so this method is also unsatisfactory. More importantly, such a withdrawal rate leads to the formation of single crystal silicon with a high concentration of self-interstitial atoms. Such high concentrations, in turn, lead to the formation of agglomerated self-interstitial defects and all the problems associated with such defects.

A second approach to addressing the problem of agglomerated intrinsic point defects involves focusing on the dissolution or disappearance of the agglomerated intrinsic point defects after their formation. This is typically done by high temperature heat treatment of silicon in wafer form. For example, Fusegawa et al., In European Patent Application No. 503816A1, have grown a silicon ingot at a growth rate greater than 0.8 mm / min and prepared a wafer 1 sliced from the ingot.
It is proposed to perform a heat treatment at a temperature of 150 ° C. to 1280 ° C. to reduce the defect density in a thin region near the wafer surface. The specific processing required will depend on the concentration and location of agglomerated intrinsic point defects on the wafer. Different wafers cut from crystals that do not have such a uniform axial concentration of defects require different post-growth processing conditions. Moreover, such wafer heat treatments are relatively costly, can introduce metallic impurities into silicon wafers, and are not generally effective in defects associated with all types of crystals.

A third method to address the problem of agglomerated intrinsic point defects is the epitaxial deposition of a thin crystalline layer of silicon on the surface of a single crystal silicon wafer. This method provides a single crystal silicon wafer having a surface that is substantially free of agglomerated intrinsic point defects.
However, epitaxial deposition adds significantly to the cost of the wafer.

DISCLOSURE OF THE INVENTION In consideration of these circumstances, there is a demand for a method for producing single crystal silicon that prevents the formation of agglomerated intrinsic point defects by suppressing the agglutination reaction that causes them. Methods that act to suppress the agglutination reaction, rather than simply limiting the rate at which such defects are formed, or eliminating some of the defects after they are formed, include agglomerated intrinsic point defects. A silicon substrate is provided which is free or substantially free. Such a method also provides single crystal silicon wafers with epi-like yield potential without the high costs associated with epitaxial methods.

It is known that silicon single crystal ingots can be grown that are substantially free of defects caused by the aggregation of intrinsic point defects (eg, PCT / US98 / 073).
56 and PCT / US8 / 07304). The major mechanism of inhibition of agglutination is radial out-diffusion of intrinsic point defects. When sufficient time is given at a crystal temperature higher than the temperature T _{A at} which the agglutination reaction occurs, self-interstitial atoms and vacancies associate with each other and disappear, or diffuse into sinks on the surface of the ingot. .

The silicon self-interstitials have a temperature close to the solidification temperature of silicon, that is, about 1410.
It is believed to be very mobile at ° C. However, this mobility decreases as the temperature of the single crystal silicon ingot decreases. Generally, the diffusion rates of self-interstitial atoms are such that they are below about 700 ° C, perhaps 800 ° C, 9
At temperatures as high as 00 ° C, 1000 ° C, or even 1050 ° C, the rates are fairly slow such that they are essentially immobile over commercial practical hours.

In this connection, the temperature at which the self-interstitial agglomeration reaction takes place theoretically extends over a wide temperature range, but in practice this range is relatively narrow for typical Czochralski-grown silicon. It should be noted that this is considered. This is a result of the relatively narrow range of initial self-interstitial concentration that is commonly obtained in silicon grown by the Czochralski method. Thus, in general, a self-interstitial agglomeration reaction is at a temperature (T _A ) of about 1100 ° C. to about 800 ° C., typically about 1050 ° C.
It can occur at temperatures of ° C.

By adjusting the cooling rate of the ingot within the temperature range in which the self-interstitial atoms are considered to be mobile, the self-interstitial atoms are dominated by sinks or vacancies located at the crystal surface where they can disappear. The time to diffuse into the area can be longer. Therefore, at a temperature low enough to prevent supersaturation of self-interstitials (ie, concentrations above the solubility limit) from occurring at temperatures at which the self-interstitials are sufficiently mobile to agglomerate. The concentration of such self-interstitials can be suppressed. The same principle applies to silicon vacancies. However, the relative immobility of the vacancies makes their outdiffusion more difficult.

While it is possible to produce single crystal ingots that do not have agglomerated micro-defects in existing crystal pullers, there are many conflicting conditions in the operation of crystal pullers and ingots. There is a need to make compromises that significantly impact the commercial viability of defect-free single crystal ingot production. The growth of a single crystal silicon ingot is shown schematically in FIG. Silicon solidifies from the melt into an ingot at about 1410 ° C., then continuously cooled. At some position along the length L (T _A ) of the ingot above the melt surface, the ingot passes the isotherm T _A (eg 1050 ° C.) at which the agglutination reaction occurs. The ingot passes through this point as it grows.

Basically, the growing defect-free ingot has a temperature T _A (
In order to obtain a sufficiently long residence time of the ingot at a temperature higher than (for example, 1050 ° C.), it is necessary to induce the temperature distribution in the hot zone to allow the out-diffusion of the intrinsic point defects. Maximizing the residence time of the ingot axial segment at temperatures above T _A requires slower pull rates. However, slowing the pulling rate significantly reduces the throughput of the crystal puller.

The dwell time at temperatures above T _A required for each axial segment of the ingot can be reduced somewhat by growing the crystal so that the self-interstitial intrinsic point defects dominate. it can. Self-interstitial defects are much more mobile than vacancy defects. It is also necessary to minimize the initial concentration of defects.
However, in order to minimize the number of defects, the pulling rate must be maximized within the interstitial growth conditions.

In order to produce a single crystal ingot that is substantially free of agglomerated microdefects along its entire length, each axial segment along the entire usable length of the ingot allows for the outward diffusion of intrinsic point defects. It must pass through T _A only after dwelling above T _{A for the} required time. Therefore, the same relatively slow pull rate must be maintained even while the unusable end cone of the ingot is being formed. Furthermore, even after the ingot has been formed, such that the lower end of the usable constant diameter portion of the ingot has sufficient residence time at temperatures above T _A.
The ingot must be pulled at a similar slow rate.

The antagonism between pulling rate and residence time required for out-diffusion of intrinsic point defects becomes more severe as the diameter of grown crystals increases. As the diameter of the ingot increases, so does the number of defects and the distance that the defects must spread to the surface of the ingot.

Furthermore, minimizing the time for outdiffusion of self-interstitial atoms is preferable for minimizing the radial change in the initial interstitial atom concentration. This is done by minimizing the radial variation of the axial temperature gradient G _O (r). In order to minimize the radial variation of the axial temperature gradient, it is preferable to minimize the average value of the axial temperature gradient G _O in the ingot on the surface of the silicon melt. However, it is preferable to minimize the average value of G _O in order to maximize the pulling rate that achieves the interstitial growth conditions.

Indeed, in order to produce a single crystal silicon ingot that is substantially free of agglomerated intrinsic point defects, very tight process control must be maintained in the operation of the crystal puller. Moreover, the throughput of the crystal puller is significantly reduced. Therefore, there is currently a need for a method of growing single crystal ingots without agglomerated intrinsic point defects that decouples or substantially decouples the operation of the crystal puller from the conditions required to diffuse the intrinsic point defects. ing.

Some objects and features of the present invention provide a method of manufacturing a single crystal silicon ingot that is substantially free of agglomerated intrinsic point defects over the entire usable length of the ingot; Providing such a method which does not substantially reduce the quantity; providing such a method which decouples the operating conditions of the crystal puller from the manufacturing conditions of the defect-free ingot; limits take-off speed for producing ingot, is substantially reduced, to provide such a method: and, the crystal withdrawal device, the limit of the average axial temperature gradient G _O, substantially reduce its It is to provide such a method.

Briefly, therefore, the present invention is directed to a method of making a single crystal silicon ingot having a seed-cone, an end cone, and a constant diameter portion between the seed cone and the end cone. The ingot is grown from silicon melt by the Czochralski method. Generally, the method grows an ingot from a silicon melt and, while the ingot is growing, the intrinsic point in the ingot is such that at least the constant diameter portion of the ingot is substantially free of agglomerated intrinsic point defects. _A method comprising adjusting the temperature of the ingot so that it does not cool any part of the ingot below the temperature T _{A at} which defect agglomeration occurs.

The present invention also relates to a method of making a single crystal silicon ingot having a seed cone, an end cone, and a constant diameter portion between the seed cone and the end cone. The ingot is grown from the silicon melt by the Czochralski method with a crystal pulling apparatus. The crystal puller has a lower growth chamber and an upper growth chamber, and the method lowers the seed crystal into contact with the silicon melt loaded in the growth chamber of the crystal puller to melt the seed crystal. And freezing the silicon from the melt to form a single crystal silicon ingot. The fully formed ingot is raised into the take-off chamber and then the take-off chamber is isolated from the growth chamber while maintaining the temperature of the take-up chamber above the temperature T _{A at which} agglomeration of intrinsic point defects in the ingot occurs. .

Other objects and features of the invention will in part be obvious and will in part appear below.

In the drawings, corresponding reference numerals indicate corresponding parts throughout the drawings.

Referring to the drawings, and in particular to FIG. 3, an apparatus for carrying out the method of the present invention having a crystal pulling apparatus, generally indicated at 10, is shown in schematic form. The crystal take-up device is
It has a growth chamber 12 having a crucible 14 containing a silicon melt M. Crucible 1
4 is rotatably arranged in the growth chamber 12 as is conventional. The crystal puller 10 also has a conventional heater and heat insulator (not shown) that heats the silicon in the crucible 14 to create and maintain its molten state. The crystal puller 10 is further located above the growth chamber 12 and can be opened towards the growth chamber to receive a single crystal silicon ingot I grown from the molten silicon M,
It also comprises a take-off chamber 16. The take-up chamber 16 has a winch mechanism 18 for raising and lowering a take-up wire 20 having a seed crystal chuck 22 at its end. Alternatively, a take-up mechanism (using a shaft rather than take-up wire (
(Not shown) can also be used. The chuck 22 holds a seed crystal (not shown) used to start the formation of the ingot I by the Szochralski method.

The collection chamber 16 has a valve 24 that closes the collection chamber 12 with respect to the growth chamber 12.
Similarly, the growth chamber 12 has its own valve 2 which closes the growth chamber 16 with respect to the take-off chamber 16.
Have six. The take-off chamber is removably arranged on the growth chamber 12 so that the entire take-off chamber 16 can be removed from the growth chamber. The device also has another take-off chamber 16 '(corresponding parts of the other take-off chamber 16' are indicated by the same numbers as the take-off chamber 16 followed by a "'"). Another take-up chamber 16 'is located above the growth chamber 12 and can be used to grow another ingot I'. However, as shown in FIG. 3, one ingot I is grown in the take-off chamber 16 while another ingot I ′ is kept in the other take-off chamber 16 ′ at a position remote from the growth chamber 12. To be done.

A method implemented in accordance with the principles of the present invention will be described with respect to the apparatus shown in FIG.
As will be appreciated, this method can also be performed by other devices, some examples of which are described below. For example, solid polysilicon is placed in the crucible 14 and energy is applied to the heater to melt the silicon to form the silicon melt M, thereby first preparing the crystal pulling apparatus 10 in a conventional manner. The winch mechanism 18 is actuated to unwind the draw wire 20 and lower the chuck 22, whereby the seed crystal contacts the surface of the melt. Both the crucible 14 and the take-up wire 20 rotate about a vertical axis. As the seed crystal begins to melt, the winch mechanism 18 operates to slowly wind the take-up wire 20 and pull the seed crystal from the melt. Silicon from the melt M solidifies on the seed crystal in the single crystal lattice, whereby the ingot I begins to form.

The ingot initially has a seed cone SC which is increased to a point corresponding to the desired ingot diameter (generally somewhat larger than the desired diameter of the semiconductor wafer ultimately formed from the ingot). Has a diameter that The constant diameter portion CD is grown by adjusting the take-up speed and heating the ingot I. When the constant-diameter portion CD reaches the desired length, the melt M ingot I
To form the end cone EC '(shown only in another ingot I'). This length is limited by the shape of the crystal puller 10. The end cone EC 'is also formed by adjusting (ie, generally increasing) the ingot pulling rate and applying heat. After being separated from the melt M, the ingot I is completely pulled up into the take-up chamber 16.

According to the method of the present invention, the temperature of the ingot I is maintained above the temperature T _{A at} which the intrinsic point defects become supersaturated and aggregate during the growth of the ingot. In particular, no portion of ingot I is allowed to cool to temperature T _A during crystal growth. Therefore, unlike the example of the conventional Czochralski method shown in FIG. 1, the ingot I never passes through the isotherm T _A during growth. The limitation in the production of single crystal silicon previously caused by the presence of the isotherm at T _A is eliminated by the method of the present invention. It is believed that adjustment of the cooling of the ingot I can be done by heat shields, application of heat, or some combination of the two. In the illustrated embodiment, a heater 30 (shown diagrammatically in FIG. 3) that applies heat to the ingot as it approaches and enters the collection chamber is attached to the collection chamber 16. .

At a predetermined concentration in the ingot below the solubility limit required for the aggregation of the intrinsic point defects to occur, the ingot I is brought to a temperature above T _A for a time selected to allow the intrinsic point defects to diffuse out. Maintained. The time required for outdiffusion of intrinsic point defects (discussed in detail below) is generally significantly longer than the standard cycle time of crystal puller 10. To this end, the method of the present invention removes ingot I from the location of crystal puller 10 in a semiconductor fabrication facility so that the crystal puller is recycled independent of the thermal conditions of the ingot. It also includes Ingot I is maintained at a temperature above T _A during or after removal from crystal puller 10.

In the embodiment shown in FIG. 3, removal of the ingot I from the crystal pulling apparatus 10 includes completely pulling the grown crystal into the crystal pulling chamber 16. The valve 24 of the take-off chamber 16 and the valve 26 of the growth chamber 12 are closed, isolating them from each other and from the surrounding environment. Next, another take-off chamber 16 in FIG.
Separating the take-off chamber 16 from the growth chamber 12, as indicated by the position '
Move and release. Until the time when sufficient outward diffusion of the intrinsic point defect occurs, the collection chamber 16
Keeps the ingot I above T _A. The ingot I is then cooled to ambient temperature and removed from the take-off chamber 16 for further processing into semiconductor wafers.

The heater of the growth chamber 12 is turned off, which cools the growth chamber to ambient temperature. The growth chamber can then be opened, thereby removing the crucible 14 and replacing it with another crystal. The solid polysilicon held in the crucible 14 is melted to form a new melt. At a suitable time after removing the take-off chamber 16, another take-off chamber 16 ′ (first removing the ingot I ′ held therein) is moved to a predetermined position in the growth chamber 12. The take-up chamber 16 'is attached to the growth chamber 12 and the take-off chamber and growth chamber valves 24' and 26 are opened to grow another single crystal silicon ingot I '.

The total time that the ingot I must be maintained above T _A depends on the initial concentration of intrinsic point defects, the type of intrinsic point defects predominant in the ingot, and the diameter of the grown ingot. Referring now to FIG. 2, the concentration of both types of intrinsic point defects is plotted against the ratio of the withdrawal rate v to the average instantaneous axial temperature gradient G _O in the ingot I at the melt surface. v low in the ratio of / G _O, self-interstitial intrinsic point defects [I] is dominant in the high specific, vacancies [V] is predominant. However, v / G _O critical ratio (a critical ra
It can be seen in tio) that the concentrations of both types of intrinsic point defects are minimized. At present, this ratio is considered to be approximately 2.1 × 10 ⁻⁵ cm ² / sK. It is preferred that v ratio of / G _O to keep close to the critical value, but over the entire growth process of the ingot I, especially in the seed corn and end cone end, it is difficult to do this. The feature of the present invention is that the growth of the ingot I depends on the v / G _O ratio because the outward diffusion of the intrinsic point defects is possible according to the present invention without significantly affecting the cycle time of the crystal pulling apparatus 10. It is a low dependency.

Preferably, in the ingot I grown by the method of the present invention, self-interstitial type intrinsic point defects are predominant. The self-interstitial defect [I] is more mobile than the vacancy defect [V]. Radial outdiffusion of self-interstitials occurs about 10 times faster than the outdiffusion of vacancies. In other words, it takes ten times as long as the time required for outdiffusion of the interstitial atoms having the same concentration in the ingot in which the interstitial atoms are dominant in order to outdiffuse the ingot in which the vacancies are dominant. As a result, v / G _O ratio is, in significant part of the growth of the ingot I, is kept below the critical value, thereby preferably to self-interstitial defects are dominant. Not surprisingly, v / G _O varies across the radius of the ingot I, so that there is a radial variation in the concentration and type of defects in the ingot. However, as long as the self-interstitial defects are predominant enough to recombine with vacancies during outdiffusion, v / G _o for the portion of ingot I is allowed to move to the vacancy predominant region. , Thereby eliminating both defects and keeping their concentration below the solubility limit.

However, it should be understood that the method of the present invention may also be used with pore-dominant materials.
Generally, the void-dominant material, if present, lies at the axial center of the ingot and extends from the center to the edge of the ingot, depending on the crystal growth conditions. When the vacancy-dominant material does not extend from the center to the edge, the core of the vacancy-dominant material is surrounded by an annular ring of interstitial-dominant material. Due to the slow mobility of vacancies in the silicon lattice (compared to silicon self-interstitials), the necessary relaxation of the vacancies system (ie suppression of vacancies concentration) by outdiffusion to the surface of vacancies is possible. It takes a relatively long time to obtain. Therefore, in one embodiment of the present invention,
The time required to suppress the vacancy concentration is reduced by injecting silicon self-interstitials into the ingot that diffuse into the existing vacancies and recombine to eliminate them. In this embodiment, silicon self-interstitials can be implanted by oxidizing the surface of the ingot while maintaining the ingot above the temperature at which the agglomeration reaction occurs. Oxidation is performed, for example, by exposing the ingot to an oxidizing atmosphere (eg, an atmosphere containing oxygen or steam, but not, but preferably substantially, an atmosphere of oxygen or steam during its maintenance time). be able to. The oxide film has a thickness of about several microns (for example, 3, 4, 5 microns), or even 10 microns or more,
Can grow Growth in one cycle, two cycles, or more cycles because the thickness of the oxide film affects the oxidation rate (which in turn affects the rate at which silicon self-interstitial atoms are implanted). It is advantageous to strip off the formed oxide film (for example with hydrogen or HF vapor) and then re-oxidize the crystal surface during the holding time.

The diameter of the growing ingot I depends on the time it takes for the diffusion of the intrinsic point defect, simply because the intrinsic point defect has to travel a longer radial distance as the diameter of the ingot increases. Affect. The time required for diffusion is
It is proportional to the square of the radius of the ingot. Therefore, the constant diameter part of the ingot is about 15
When it is 0 mm, the temperature of the ingot is higher than T _A (ie, 1050 ° C., 100
The total residence time at 0 ° C, or 900 ° C) was found to be at least about 10 hours, preferably at least about 12 hours, more preferably at least about 15 hours. Thus, the total time a 200 mm ingot dwells above T _A in a similar system is at least about 22 hours, preferably at least about 2 hours.
5 hours, more preferably at least about 30 hours, while the total time the 300 mm ingot dwells above T _{A in} similar systems is at least about 48 hours, preferably at least about 60 hours, more preferably At least about 75
It's time. The exact time of out-diffusion other than that, without departing from the scope of the present invention,
It is understood that it is other than the one described.

Referring now to FIG. 4, the method of the second embodiment is shown. This method is similar to the method of the first embodiment, except that the pulling chamber 116 of the crystal pulling device 110 is not removed from the growth chamber 112. In the second embodiment, the collection chamber 116 is modified to open to the holding chamber 140 next to the collection chamber. For convenience of disclosure of this method, the holding chamber 140 shall not form part of the crystal puller 110, even when it is physically attached to the crystal puller. After the ingot I has grown sufficiently and has been pulled up into the take-up chamber 116, a door (not shown) that separates the take-up chamber from the holding chamber 140 is opened, the ingot is transferred to the holding chamber, and the T Keep the ingot above _A temperature. Track 1 that carries the ingot I to the holding chamber 140
42 is shown. In the embodiment shown, the holding chamber 140 includes the heater 1
Has 44. Next, a thermal lock (not shown) that allows cooling and removal of the ingot from the holding chamber 140 without compromising the thermal environment of the holding chamber,
Move the ingot. Meanwhile, the door separating the take-off chamber 116 from the holding chamber 140 can be closed. In order to grow another single crystal ingot (not shown), another winch mechanism and take-up wire (not shown) are moved into place in the take-off chamber 116.

A third embodiment of the method of the present invention is shown in FIG. This method is very similar to the method shown in FIG. 4, since the ingot I is moved to the holding chamber 240 without moving the take-off chamber 216. The main difference is that the holding chamber 240 is the collecting chamber 21
It is to be placed above 6 rather than next to it. Again, another ingot I in the crystal pulling apparatus 10 will be used regardless of the thermal conditions of the already grown ingot I.
'Another winch mechanism 218' and take-up wire 22 so that growth of'takes place.
Move 0 '.

It is also noted that in the above-described embodiment of the method of the present invention in which the grown ingot is held in the growth chamber to provide sufficient time for out-diffusion, it is preferred that the take-off chamber have a non-uniform thermal profile. Should. In other words, when at least some of the grown ingot is cooled or removed from the growth chamber because it has been cooled to a temperature below the temperature at which agglomerated intrinsic point defects are formed (T _A ), It is not necessary to keep this part above T _A. In fact, if the temperature is too high (ie above about 1175 ° C., 1200 ° C. or higher), the concentration of intrinsic point defects will again be above the solubility limit or critical concentration as a result of diffusion,
It is preferred that the temperature profile of this part of the crystal does not exceed T _A. However, the temperature in this region should not be too high, while the temperature in other parts of the ingot is
The temperature must be maintained high enough so that no agglomeration occurs.

When a heterogeneous heat profile is used, it is preferred that the temperature be gradually increased from the seed end to the tail end, generally in the range of about 1000 ° C to about 1200 ° C, preferably about 1050 ° C to about 1175 ° C. Goes up. The axial position of the ingot, which has a temperature above T _A , is then cooled by the method of the invention, preferably until the temperature profile is uniform. The ingot is then further cooled and removed for further processing, as is customary in the art.

It should be noted that although a non-uniform temperature profile is preferred, a uniform temperature profile can also be used. However, if a uniform profile is used, the temperature must be sufficiently higher than T _A to prevent agglomeration from occurring, but the region previously cooled to a temperature below T _A will once again reach critical supersaturation. It should not be so high as described above (described above). Therefore, the temperature is preferably about 1125 ° C to about 120 ° C.
It is 0 degreeC, More preferably, it is about 1150 degreeC to about 1175 degreeC. Once the ingot enters the chamber, the temperature of this uniform profile is reduced according to the present invention, cooling the ingot below T _A. The ingot is then further cooled and removed for additional processing, as is conventional in the art.

[0045]   The following example illustrates the invention.

(Example) The velocity indicated by a dashed line in FIG. 6A (hereinafter, “defect-free”)
Two 200 mm crystal ingots were grown in a crystal puller capable of producing a material with no agglomerated intrinsic point defects when the ingots were grown according to the growth rate curve). The two crystals were grown at similar target growth rates, shown as continuous lines in FIG. 6A, and the growth rates were recorded as standardized growth rates (ie, generally as a ratio of actual growth rate to critical growth rate). The actual growth rate relative to the critical growth rate, expressed). As shown, ingot first, "defect free"
Grow at a rate faster than the growth rate curve for a period of time, then at a rate slower than the "defect-free" growth rate curve for a period of time, then again at a rate faster than the "defect-free" growth rate curve. Grow over time. After the growth of the ingot, the first ingot (87GEX) was naturally cooled in the crystal growth chamber. However, the second ingot (87GEW) does not cool naturally in the crystal growth chamber; instead, after the growth of the ingot, leave the heater in the hot zone of the crystal pulling device on and keep the ingot in the pulling chamber. For 30 minutes; the temperature profile during this time is about 10 mm over the area of the ingot about 400 mm or more from the seed end.
Maintaining a temperature higher than 50 ° C, the area less than about 400mm from the seed end is about 1
The temperature profile was such that the temperature was maintained below 050 ° C.

The ingot was sliced longitudinally along a central axis parallel to the direction of growth and then divided into several parts each having a thickness of about 2 mm. Suitable for using a copper decoration method to artificially contaminate the set of longitudinal parts that make up each ingot from the seed to the tail with copper to dissolve high concentrations of copper in the parts. It heated on heating conditions. After this heat treatment, the sample was rapidly cooled, during which copper diffused out or precipitated at sites where there were oxide clusters or cohesive intrusion defects. Standard defect description etching (standa
After rd defect delineating etch) the samples were visually inspected for the presence of precipitated impurities, those areas without such precipitated impurities corresponded to areas without agglomerated interstitial defects. Next, a picture of each crystal part was taken and the pictures were combined to show the results from seed to tail end of each crystal. A set of photographs of the first naturally cooled ingot (87GEX) is shown in FIG. 6B and the second sustaining ingot (87GEX).
A set of photographs of EW) is shown in FIG. 6C.

Referring now to FIGS. 6A, 6B, and 6C, a naturally cooled ingot (87G
EX) has an agglomerated vacancy defect in 0 to about 393 mm, no agglomerated intrinsic point defect in about 393 mm to about 453 mm, and an agglomerated intrinsic point defect in about 435 mm to about 513 mm, about 5
It can be seen that there is no agglomerated intrinsic point defect at 13 mm to about 557 mm and agglomerated vacancy defect at 557 mm to the crystal end. These correspond to regions above, below, or below the defect-free growth conditions of this hot zone. Maintenance ingot (87GEW
) Has agglomeration vacancy defects at 0 to about 395 mm, no agglomeration intrinsic point defects at about 395 mm to about 584 mm, and agglomeration vacancy defects at about 584 mm to the crystal end. Therefore, 2
The largest difference between the two ingots occurs in the region of about 435 mm to about 513 mm where the naturally cooled ingot (87GEX) has agglomerated intrinsic point defects, while the maintenance ingot (87GEW) has no agglomerated intrinsic point defects. During the maintenance time, the concentration of self-interstitial silicon atoms in the maintenance ingot (87GEW) is suppressed by the additional diffusion of self-interstitial atoms to the surface of the ingot and the vacancy-dominated region,
Therefore, critical supersaturation and agglomeration reactions of interstitial atoms are prevented after crystal solidification.
However, in the naturally cooled ingot, sufficient time is not given for additional diffusion to the surface of the ingot and the vacancy-dominated region, and as a result, the system becomes a critical supersaturation of silicon self-interstitials and the agglomeration reaction is Occur.

Thus, these ingots show that virtually any amount of silicon self-interstitials can diffuse out to the surface if given sufficient time and sufficiently high temperature.

Further, the “defect-free” growth rate curve shown in FIG. 6A is a range of crystal growth rates that gives a material having no agglomerated intrinsic point defects under natural cooling conditions in the configuration of this crystal pulling apparatus. include. Even under the natural cooling condition of this hot zone arrangement, the range of the crystal growth rate is between the growth rate at which agglomerated vacancy defects are formed (P _V ) and the growth rate at which agglomerated intrinsic point defects are formed (P _I ). Is present and this range is at least ± 5% of the average of P _V and P _I. About 1050 ℃ of grown crystal
This range is further increased when the residence time at higher temperatures is increased,
The range is, for example, at least ± 7.5%, at least ± 10%, or at least ± 15% of the average of P _V and P _I (eg, crystalline 87GEW).
, The retention was long enough so that P _I was not reached and thus the P _I of this crystal was slower than the lowest take-up rate reached). The results are shown in Table I below.

[0051]

[Table 1] Table I

Increasing the window size (or the change in draw rate allowed for defect-free growth) is substantially limited to slower draw rates (ie, vacancies to interstitial dominance). Values below the critical v / G for the material (and limited to some small delta to take into account the interstitial annihilation of vacancies). That is, the effect is highest for interstitial-dominant materials, where silicon self-interstitials are faster diffusion elements than vacancies. In other words, the window opens more quickly for slow take-off speeds. In principle, the window of allowable withdrawal rate changes towards faster withdrawal rates (greater than the critical v / G value and some small delta) into the vacancy-dominant material is also a factor when the vacancy diffuses to the crystal surface. About 10
Increased residence time at temperatures above 50 ° C. opens towards higher take-off rates (pore-dominant material), but this takes considerably longer.

[0053]   Axial temperature gradient G for a given crystal puller and hot zone configuration _O Compare as occurs in the transition ranges that occur here
It is almost constant over a very short distance. As a result, the change in crystal growth rate is_ORatio of
An exemplary change, and therefore a proportional change in the initial concentration of vacancies and silicon self-interstitials.
Lead to conversion. However, in general, v / G in the center ingot_O
The value of is the most critical value because it is the furthest away from the surface. Obey
And the pulling effect obtained by the increased residence time at temperatures above about 1000 ° C.
The increase in the change in the take-up speed is v / G_OA corresponding change in is along the radius of the crystal
The results of this example show that it can happen at any point.
ing. In other words, v / G_ORadial changes of are not relevant, so an example
For example, v / G at the center of the ingot_O10%, 15% or more of the value of
Can be exceeded (at any radial position).

In view of the above, it will be seen that the several objects of the invention are achieved. All changes contained in the above disclosure are exemplary and not intended to be limiting, as various changes may be made to the above configurations and methods without departing from the scope of the present invention. It should be understood that there is no.

[Brief description of drawings]

FIG. 1 is a schematic diagram showing existing single crystal silicon ingot growth showing the ingot passing through the isotherm during the agglutination reaction during growth.

[2] the initial concentration of self-interstitials [I] and vacancies [V], v / G _O
[V is takeoff speed, the G _O is the average axial temperature gradient is a graph showing an example of changing the value of.

FIG. 3 is a schematic view of a crystal pulling device that allows removal of the pulling chamber portion of the crystal pulling device and replacement with another pulling chamber.

FIG. 4 is a schematic view of a crystal pulling apparatus that allows the ingot to be removed from the pulling chamber and transferred to a holding chamber located next to the pulling chamber.

FIG. 5 is a schematic view of a crystal pulling apparatus that allows an ingot to be removed from the pulling chamber and transferred to a holding chamber located substantially above the pulling chamber.

FIG. 6A is a graph showing the relationship of standardized growth rate to crystal length, as shown in the examples.

FIG. 6B is a series of photographs of axial cuts of a segment of an ingot from shoulder to initiation of end-cone growth after copper decoration and defect delineation etching as shown in the examples.

FIG. 6C is a series of photographs of an axial cut of a segment of an ingot from seed cone to end cone after copper decoration and defect delineation etching as shown in the examples.

[Procedure for Amendment] Submission for translation of Article 34 Amendment of Patent Cooperation Treaty

[Submission date] July 10, 2000 (2000.7.10)

[Procedure Amendment 1]

[Document name to be amended] Statement

[Name of item to be amended] Claims

[Correction method] Change

[Contents of correction]

[Claims]

─────────────────────────────────────────────────── ─── Continued front page    (31) Priority claim number 60 / 117,623 (32) Priority date January 28, 1999 (January 28, 1999) (33) Priority claiming countries United States (US) (81) Designated countries EP (AT, BE, CH, CY, DE, DK, ES, FI, FR, GB, GR, IE, I T, LU, MC, NL, PT, SE), CN, JP, K R, SG

Claims (31)

[Claims] 1. A method of manufacturing a single crystal silicon ingot having a seed cone, an end cone, and a constant diameter portion between the seed cone and the end cone, the method comprising:
The ingot is grown from a silicon melt by the Czochralski method, the method growing the ingot from the silicon melt, and such that at least the constant diameter portion of the ingot is substantially free of agglomerated intrinsic point defects. _A method comprising adjusting the temperature of the ingot so that it does not cool any part of the ingot to a temperature below the temperature T _{A at which} agglomeration of intrinsic point defects occurs in the ingot while it grows. 2. The ingot is heated to a temperature above T _A for a period of time selected to allow outdiffusion of the intrinsic point defects to obtain a concentration below the solubility limit required for the aggregation of the intrinsic point defects. The method of claim 1, wherein the method is maintained. 3. The ingot is T _A for a period of time following the growth of the ingot.
The method of claim 2, wherein the method is maintained at a higher temperature. 4. The method of claim 3, wherein T _A is below the freezing temperature of silicon and above about 1050 ° C. 5. The method of claim 3, wherein T _A is below the freezing temperature of silicon and above about 900 ° C. 6. The method of claim 3, wherein the ingot is grown by adjusting the growth rate v and the average axial temperature gradient G _O such that the ingot has a predominance of self-interstitial intrinsic point defects. 7. The method of claim 3, wherein the ratio v / G _O is adjusted to be less than about 2.1 × 10 ⁻⁵ cm ² / s · K. 8. The ingot having a constant diameter portion having a diameter of about 150 mm, and the time from initiation of growth of the ingot at which the ingot is maintained above T _A is at least about 12 hours. the method of. 9. The ingot has a constant diameter portion having a diameter of about 200 mm and the time from initiation of growth of the ingot at which the ingot is maintained above T _A is at least about 22 hours. the method of. 10. The time from the start of growth of the ingot, wherein the diameter of the constant diameter portion of the ingot is about 300 mm, and the ingot is maintained at a temperature higher than T _A ,
The method of claim 3, wherein the method is at least about 48 hours. 11. The method of claim 1, which is performed in a crystal puller at a location in a semiconductor manufacturing facility, the method removing the ingot from the location while maintaining the ingot at a temperature above T _A. The method further comprising the step of removing so that the crystal puller is cooled and restarted for growth of another single crystal silicon ingot. 12. The crystal pulling apparatus has a lower growth chamber and an upper pulling chamber, and the step of removing the ingot isolates the pulling chamber from the growth chamber, separates the pulling chamber from the growth chamber, and removes the pulling chamber. The method of claim 11, comprising moving away from the growth chamber. 13. The method of claim 12, further comprising moving another take-off chamber to a position above the growth chamber and attaching the another take-off chamber to the growth chamber. 14. The crystal pulling apparatus has a lower growth chamber and an upper pulling chamber, and the step of removing the ingot comprises moving the ingot from the pulling chamber to a holding chamber adjacent to the pulling chamber. The method of claim 11, wherein the holding chamber is heated to maintain the ingot at a temperature above T _A. 15. The method of claim 1 wherein the ingot is exposed to an oxidizing atmosphere while maintaining the ingot at a temperature above the temperature T _{A at which} aggregation of intrinsic point defects occurs in the ingot. 16. The ingot is exposed to at least one cycle, the ingot being exposed to an oxidizing atmosphere in a first stage of the cycle, while maintaining the ingot above a temperature T _{A at which} agglomeration of the intrinsic point defects in the ingot occurs. Then
In the second stage of the cycle, dissolve the silicon dioxide from the surface of the ingot,
16. The method of claim 15, wherein the ingot is exposed to the atmosphere that would otherwise be removed. 17. A method for producing a single crystal silicon ingot having a seed cone, an end cone, and a constant diameter portion between the seed cone and the end cone, wherein the ingot is made of silicon in a crystal pulling apparatus by the Czochralski method. Grown from a melt, the crystal pulling device having a lower growth chamber and an upper pulling chamber, the method dropping a seed crystal into contact with a silicon melt located in the growth chamber of the crystal pulling device; A crystal is pulled from the melt to solidify the silicon from the melt to form a single crystal silicon ingot; a fully formed ingot is pulled into the take-up chamber; the take-up chamber is isolated from the growth chamber; and the intrinsic point in the ingot to maintain the temperature of the take-off chamber to a temperature higher than the temperature T _a of agglomerated defects occurs ; Process that comprises. 18. An ingot from T _A for a period of time selected to allow outdiffusion of the intrinsic point defects to obtain a concentration below the solubility limit required for the aggregation of the intrinsic point defects to occur. 18. The method of claim 17, which is maintained at an elevated temperature. 19. The method of claim 18, wherein T _A is below the freezing temperature of silicon and above about 1050 ° C. 20. The method of claim 18, wherein T _A is below the freezing temperature of silicon and above about 900 ° C. 21. The method of claim 18, wherein the ingot is grown by adjusting the growth rate v and the average axial temperature gradient G _O such that the ingot has a predominance of self-interstitial intrinsic point defects. . 22. The method of claim 18, wherein the v / G _O ratio is adjusted to be less than about 2.1 × 10 ⁻⁵ cm ² / s · K. 23. The ingot having a constant diameter portion having a diameter of about 150 mm, wherein the ingot is maintained at a temperature above T _A , the time from the start of ingot growth being at least about 12 hours. The method described. 24. The ingot having a constant diameter portion having a diameter of about 200 mm, wherein the time from initiation of ingot growth at which the ingot is maintained at a temperature above T _A is at least about 22 hours. The method described. 25. The method of claim 18, wherein the constant diameter portion of the ingot has a diameter of about 300 mm and the time at which the ingot is maintained above T _A from the beginning of growth of the ingot is at least about 48 hours. The method described. 26. The method of claim 17, wherein the crystal puller is located at a location in a semiconductor manufacturing facility, the method maintaining the ingot at a temperature above T _A. The method further comprising removing from the location so that the crystal puller is cooled and restarted for growth of another single crystal silicon ingot. 27. The step of removing the ingot comprises separating the haul chamber from the growth chamber and moving the haul chamber away from the growth chamber.
The method described in. 28. The method further comprising moving the another take-up chamber to a position above the growth chamber and attaching the another take-off chamber to the growth chamber.
7. The method according to 7. 29. The step of removing the ingot comprises moving the ingot from the pick-up chamber to a holding chamber adjacent to the pick-up chamber, heating the holding chamber to bring the ingot to a temperature above T _A. 27. The method of claim 26, which maintains. 30. The method of claim 17, wherein the ingot is exposed to an oxidizing atmosphere while maintaining the ingot above a temperature T _{A at which} aggregation of intrinsic point defects in the ingot occurs. 31. Exposing the ingot to at least one cycle, maintaining the ingot at a temperature above a temperature T _{A at which} agglomeration of intrinsic point defects occurs in the ingot, exposing the ingot to an oxidizing atmosphere in a first stage of the cycle. Then
In the second stage of the cycle, dissolve the silicon dioxide from the surface of the ingot,
31. The method of claim 30, wherein the ingot is exposed to the otherwise removing atmosphere.